The combination of NMR spectroscopy and molecular dynamics (MD) simulations are powerful tools in the studying of bioorganic molecules in solution. In this thesis two such studies are presented with focus on the NMR aspect. The caffeine association to sugars (D-glucose and sucrose) was investigated by NMR titrations and NOESY experiments in paper I. The observations from the NMR experiments confirmed MD simulations showing that the binding occurs by a face-to-face interaction between the aromatic surface of the caffeine and axial protons of the sugar ring. Different sugar molecules and residues have different preferences regarding which side of the sugar ring that are involved in the binding. The sucrose residues bind with only one ring face each whereas β-D-glucopyranose has two sides of similar binding probability and the α-D-glucopyranose has something in between. The MD simulations showed that the driving force of the binding is partly driven by hydration effects that favor the enthalpy of the system. A new approach to calculate NMR relaxation parameters (that is dependent on molecular motions) from computational simulations is presented in paper II. Each sugar residue is assumed to be a rigid unit connected by flexible joints in the approach, thus the name diffusive chain model (DCM). The simplified model together with a stochastic simulation approach lowers the computational cost which makes it possible to acquire long enough trajectories to the calculations of spin relaxation parameters. Two case studies with slightly different methodologies are presented. In one of them, spin relaxation parameters are reproduced for the human milk oligosaccharide LNF-1 in a feasible way by the use of Brownian dynamics.